Learning Week 10 and 11
Workshop (PHRM3301, PHRM3031)
Adam La Caze
Learning objectives for the week
LW10: Diagnostic tests and screening
- Be able to describe and calculate the properties of a test for disease: sensitivity, specificity, positive predictive value, negative predictive value
- Be able to assess the usefulness of a diagnostic or screening test
- Be able to describe the common sources of bias when assessing a screening test
LW11: How things can go wrong
- How can the pharmaceutical industry influence decisions?
- How do we treat individuals based on data from populations?
- Understanding the ways in which our methods are fallible
Extend activities
LW10
- Diagnostic tests
- Covid-19; sensitivity, specificity, positive predictive value, negative predictive value
- Troponin and acute coronary syndrome
- Deciding test cut-offs
- Disease screening
- Breast cancer screening
LW11
- Marketing-based medicine (Spielmans and Parry 2010)
- The subgroup problem
- Clopidogrel and ticagrelor
- Why most published research findings are false*
Course and teaching evaluation (PHRM3301)
Written exam
Details (PHRM3301)
- The written exam is worth 30% of the marks of the course
- It as an on-campus invigilated exam via Inspera.
- The exam will be 120 minutes with 10 minutes reading time.
- 120 marks (1 mark per minute)
- 80 marks clinical epidemiology (Adam: LW1–4, 10–11)
- 40 marks statistics (Ian: LW5–9)
- See the sample exam!
Details (PHRM3031)
- The written exam is worth 50% of the marks of the course
- It as an on-campus invigilated exam via Inspera.
- The exam will be 60 minutes with 10 minutes reading time.
- See the sample exam!
Diagnostic tests
Covid-19, RAT test properties
A company that has produced a new rapid antigen test to detect Covid-19 needs to demonstrate that it has sufficient sensitivity and specificity to pass regulatory requirements.
They conduct a study comparing the RAT outcomes to a real-time reverse-transcription polymerase chain reaction (rRT-PCR) test.
100 people with symptoms of respiratory tract infection are enrolled, 15 people are identified as Covid-positive via rRT-PCR. Of these 15 people, 14 are identified as Covid-positive according to the RAT. Three people with a negative rRT-PCR test were identified as positive according to the RAT.
Determine the sensitivity and specificity of the RAT test and the prevalence of Covid in this sample.
| Condition | No condition | Total | |
|---|---|---|---|
| Test \(+\) | 14 | 3 | 17 |
| Test \(-\) | 1 | 82 | 83 |
| Total | 15 | 85 | 100 |
\(Sn = \frac{a}{a + c} = 14/15= 93\%\)
\(Sp = \frac{d}{b + d} = 82/85 = 96\%\)
\(P = \frac{a + c}{a + b + c + d} = 15/100 = 15\%\)
Predict what will happen to the positive predictive value and negative predictive value of the test if the prevalence of Covid in the community changes.
| Condition | No condition | Total | |
|---|---|---|---|
| Test \(+\) | \(a\) | \(b\) | \(a + b\) |
| Test \(-\) | \(c\) | \(d\) | \(c + d\) |
| Total | \(a + c\) | \(b + d\) | \(a + b + c + d\) |
Disease screening
Early detection of breast cancer by mammography
Review the information provided by Harding Centre for Risk Literacy (below).
- How would you describe this data to a 50 year old woman considering screening?
- What are the uncertainties in the data? (What are the challenges in obtaining reliable data?)
- Why aren’t incidence rates and 5-year survival reported?
How things can go wrong
Marketing-based medicine: Problems and replies
Describe ways that the pharmaceutical industry can undermine evidence-based medicine
What tools are available to identify and/or mitigate these attempts?
The subgroup problem
Explain the subgroup problem.
Semaglutide and cardiovascular outcomes (Lincoff et al. 2023)
| Participants | People over 45 with BMI > 27 and established cardiocascular disease and NOT a diagnosis of diabetes |
| Intervention | Semaglutide sc once-weekly (escalating dose to 2.4 mg) |
| Comparator | Placebo |
| Outcome | Composite: Death from cardiovascular disease, nonfatal MI, nonfatal stroke |
The trial included 17 604 participants and the sample was ~ 72% male and 84% white
Ioannidis’ Why most published research is false*
Explain the following table:
